data-analytics
by markdown-viewerThe data-analytics skill creates PlantUML diagrams for data analysis workflows, including ETL, ELT, data lakes, warehouses, streaming pipelines, log analytics, and BI dashboards. It is optimized for clear source-to-destination flow, AWS analytics/database stencils, and practical data-analytics guide output—not generic software or cloud architecture diagrams.
This skill scores 78/100, which means it is a solid listing candidate for directory users. It gives enough concrete workflow guidance to help an agent trigger the right kind of output (data analytics and pipeline diagrams in PlantUML) with less guesswork than a generic prompt, though users should expect a few adoption gaps such as missing install command and limited supporting files.
- Strong triggerability: the frontmatter clearly scopes the skill to data analytics and pipeline diagrams, with explicit NOT-use guidance against general UML/cloud modeling.
- Operationally useful workflow: it gives a quick start, critical rules, and PlantUML-specific constraints like @startuml/@enduml, left-to-right flow, and async dashed links.
- Good install decision value: multiple example files cover real analytics patterns such as ETL, data lakes, warehouses, CDC, log analytics, and BI dashboards.
- No support files or install command are provided, so adoption depends mainly on SKILL.md and examples rather than executable tooling.
- The skill is narrowly specialized to AWS/MxGraph analytics stencils, so it is less useful for non-AWS analytics architectures or general diagramming.
Overview of data-analytics skill
The data-analytics skill helps you generate PlantUML diagrams for analytics systems: ETL flows, data lakes, warehouses, streaming pipelines, log analytics, and BI dashboards. It is the right fit when you need a data-analytics guide for turning a rough architecture into a clear diagram with AWS analytics and database stencils, not just a generic prompt that names components.
Use this data-analytics skill if you want fast, readable diagrams for data analysis workflows where the pipeline order matters: source, ingest, transform, store, and visualize. It is especially useful when you need to show governance, staging, cataloging, or near-real-time movement across systems.
Best fit for pipeline and warehouse diagrams
The skill is strongest when the output should communicate how data moves, not just what tools exist. That includes ETL/ELT, CDC, lakehouse-style layouts, Redshift-centered warehouses, and operational-to-analytics handoffs. If your goal is a data-analytics for Data Analysis diagram that stakeholders can skim quickly, this skill is a good match.
What makes this skill different
The repository is opinionated about diagram structure and syntax: it expects PlantUML fences, @startuml / @enduml, left-to-right flow, and mxgraph.aws4.* stencil icons. That makes the resulting diagrams more consistent than a free-form prompt, and it reduces guesswork around icon choice and layout.
When not to use it
Do not use data-analytics for general software architecture, UML class diagrams, or broad cloud infrastructure maps. If the main story is application components rather than data movement, a different skill will produce a better result and fewer corrections.
How to Use data-analytics skill
Install and verify the skill context
For a normal data-analytics install, add the skill from the repo and then inspect the top-level instruction file first:
- Install with
npx skills add markdown-viewer/skills --skill data-analytics. - Open
SKILL.mdto confirm the diagram rules. - Check the example files in
examples/before drafting your own prompt.
The skill is compact, so the examples matter more than a long rules section. They show the actual syntax patterns the model is expected to follow.
Start from the workflow, not the tool list
A strong data-analytics usage request describes the data story in stages, not as a bag of AWS services. For example, instead of “make a warehouse diagram with Redshift and Glue,” use a prompt that specifies:
- sources: RDS, S3, Kafka, DynamoDB
- ingest path: batch, streaming, CDC, or scheduled ETL
- transforms: validation, schema mapping, enrichment
- destination: S3 lake, Redshift, Athena, or OpenSearch
- consumers: dashboards, analysts, ML features, or alerts
That structure helps the skill choose the right stencils and arrows.
Read the right examples first
For the fastest ramp-up, preview these files in order:
SKILL.mdexamples/etl-pipeline.mdexamples/data-lake.mdexamples/data-warehouse.mdexamples/real-time-streaming.mdexamples/multi-source-bi.md
If your use case is specialized, also inspect examples/cdc-pipeline.md, examples/log-analytics.md, or examples/ml-feature-pipeline.md. These examples show how the data-analytics skill handles edge cases like asynchronous flow, warehouse loading, and feature engineering.
Prompting tips that improve output quality
A good prompt for this skill gives enough domain detail to prevent generic diagrams. Include the source systems, whether flow is batch or streaming, and what “done” means for the data. For example, “show daily orders from PostgreSQL into S3 Parquet, then Glue ETL into Redshift for QuickSight reporting” is much better than “draw an analytics pipeline.”
If you need a tighter result, specify the stages you want visible and the ones you want omitted. That keeps the diagram focused and avoids unnecessary boxes.
data-analytics skill FAQ
Is this only for AWS-based diagrams?
Mostly yes. The data-analytics skill is built around mxgraph.aws4.* stencils, so it is best when AWS services are part of the architecture or when you want AWS-style analytics symbols. If your stack is mostly non-AWS, the skill may still work, but the output will be less natural.
How is this different from a normal prompt?
A normal prompt can describe a pipeline, but the data-analytics skill encodes diagram syntax, flow direction, and icon conventions. That matters when you want reliable PlantUML output instead of a one-off sketch. The skill is more repeatable for data-analytics usage because it nudges the model toward consistent structure.
Is it beginner-friendly?
Yes, if you can describe your data flow in plain language. You do not need to know PlantUML deeply, but you do need to name the major stages and endpoints clearly. Beginners usually get the best results by copying one example pattern and replacing the systems with their own.
When should I choose a different skill?
Use something else if you need generic UML, app service topology, or provider-neutral cloud infrastructure. data-analytics is strongest when the primary object is the movement and transformation of data, not the deployment of applications.
How to Improve data-analytics skill
Give the skill the business outcome
The best data-analytics results come from prompts that explain why the diagram exists. Say whether the audience is an engineer, analyst, or executive, and whether the diagram must emphasize latency, governance, cost, or reporting. That changes which stages deserve visual prominence.
Include the constraints that affect the design
If the pipeline has schema drift, late-arriving events, compliance boundaries, or multiple consumers, mention that up front. Those constraints help the skill choose meaningful elements like crawlers, catalogs, staging buckets, or async arrows instead of a simplistic straight line.
Use concrete inputs and preferred shape
Stronger inputs look like this:
- “Batch ETL from Salesforce and PostgreSQL into S3, then Redshift, with a Glue crawler and data quality gate”
- “Real-time clickstream from Kinesis to Lambda enrichment, then OpenSearch and S3 archive”
- “CDC from Aurora and DynamoDB into a warehouse with staging and replay handling”
These are better than vague requests because they define the path, not just the destination.
Iterate by checking the weakest stage first
After the first diagram, review the part that most often breaks trust: source labeling, transform naming, or sink selection. If the flow is correct but too broad, narrow the prompt to a single pipeline. If the diagram is correct but too thin, add one more stage that matters operationally, such as a catalog, validation step, or BI consumer.